SAME 2013 Conference BLUETOOTH SMART LOW POWER SENSORS Atef AL NUKARI, Pascal CIAIS, Insight SiP Sophia-Antipolis, France Abstract Low power wireless sensing applications pose great challenges for hardware/software platform design, cost, performance, and flexibility. Depending on the sensor type, the technology, the sampling time and power supply characteristics, the platform may need to incorporate different features in order to operate in a low-power, energy-efficient manner. The Bluetooth Low Energy (BLE) standard, now called Bluetooth Smart, is an ideal candidate for low-power wireless sensing applications. This is due to drastically reduced maximum peak current consumption and low duty cycles with synchronized connection periods. In this paper, we present the design and performance of three different Bluetooth Smart low-power sensor devices, optimized for current consumption. The resulting sensor devices with 3V coin cell battery CR1632 have overall PCB dimensions of 18 x 29 x 6 mm 3. A complete demonstrator system will be presented using software developed in house, allowing remote sensing of three different physical parameters. Introduction Ultra Low Power (ULP) wireless devices based on the Bluetooth Low Energy standard have received much attention recently due to their low power consumption and low duty cycles. BLE devices can use standard coin cell batteries and have life-times from several months to several years. A sensing system contains much more than digital circuits. It also contains a variety of sensing devices and wireless interfaces. This poses a number of challenges for low-power hardware/software platform design, including size, cost, power consumption, connectivity, performance, and flexibility. The Bluetooth Smart sensor nodes described herein may be used in a wireless sensor network, to capture environmental information and send it back to a base station. These nodes are cheap, small and run for a year without the need to replace the battery. In this context, Insight SiP has designed several Bluetooth Smart sensor nodes, based on a Systemin-Package (SiP) module (ISP091201) [1]. These sensor nodes contain a low power SiP module that integrates both a miniature Antenna-in-Package (AiP), and all the electronic components (transceiver, quartz, SMT components) to ensure RF communication at 2.4 GHz. The node also contains a low power sensing device and the low power microprocessor LPC1114 from NXP. The resulting sensor node with 3V coin cell battery CR1632 has overall PCB dimensions of 18 x 29 x 6 mm 3 that make it ideally suited to highly space constrained applications, as presented in Figure 1. Different sensor nodes can be implemented using the same PCB with minor modifications related to the sensor. ISP091201 µprocessor Coin Cell Battery Figure 1: Bluetooth Smart sensor module: top and bottom views In this paper, the design of these Bluetooth Smart sensor nodes is explained in detail. In particular the optimization of the current consumption is clearly demonstrated. The design and performance of three different Bluetooth Smart sensor nodes will be presented, especially the low-power hardware/software platform design. The first sensor node is an accelerometer device using MMA7660FC from Freescale Semiconductor, the second one is a temperature sensor using TMP112 from Texas Instruments and the last one is an ambient light sensor using APDS-9300 from Avago Technologies. Two complementary methods for current consumption measurements will be presented, in addition to a life time calculation model. Conclusions and perspectives based on the results and comparisons between the model and measurements will be presented. The prototypes have been manufactured using high volume subcontractors. Current consumption characterization shows performance equivalent to the reference data sheets of the SiP module, the microprocessor and the sensors. SAME 2013 Conference October 2 & 3, 2013 1
Finally, a demonstrator system will be presented. A short visual presentation of potential applications of these wireless sensor nodes will be made using software developed in house. measurements presented in Figure 3 are made using an oscilloscope. Design of smart BLE sensor nodes Bluetooth Smart sensor nodes contain low power System-in-Package module ISP091201 (Insight SiP), low power host microprocessor LPC1114FHI33/302 (NXP) and low power sensing device, as presented in Figure 2. Low Power I2C Low Power uprocessor SPI Interface BLE Tx/Rx Figure 3: Operational phases of the Bluetooth Smart sensor nodes Isensor Figure 2: Schema block of Bluetooth Smart sensor modules The microprocessor uses a SPI interface to communicate with the BLE and an I2C interface to communicate with the sensing device. The specifications of these nodes depend on the sensing device used, as presented in Table 1. Table 1: Specifications of Bluetooth Smart sensor nodes Module Parameter Value Unit Power supply 3 V Coin cell Battery CR1632 Connection Interval 7.5 to 4000 ms PCB dimensions 18 x 29 x 6 mm 3 Accuracy +/- 0.5 (0 C Temperature 65) Temperature range -40 to +125 C Resolution 12 bits Light Light range Dark to bright sun Lux Orientation/ Motion IuP Coin cell battery CR1632 Resolution 16 bits Number of axes 3 Acceleration range +/- 1.5 g 9.8m/s² Current consumption/optimization The behavior of these Bluetooth Smart sensor nodes can be separated in two phases: : phase of slave-master communication or connection phase. In this phase, the BLE module communicates with the master to send the measurements read by the sensing devices and sent to BLE module by the microprocessor. Thus, the BLE module, the microprocessor and sensing device are active and consume the maximum of current. : Standby (Sleep) phase. In this phase, there is no communication between the BLE module and the master. Thus, the BLE module and the microprocessor are in Standby mode. Figure 3 illustrates the two operational phases of the Bluetooth Smart temperature node. The Ible Module BLE Autonomous According to the component data sheet, Figure 4 illustrates the principle of current drain profile for a typical BLE module ISP091201 which is in a connection event (Figure 4-a). It illustrates also the principle of current drain profile for the microprocessor LPC1114FHI33/302 (Figure 4-b) and for the sensing device (Figure 4-c). For the BLE module, each connection event consists of the following states and operations (different periods related to each connection event), as presented in Figure 4-a. The current consumption profile is related to each state and operation. The numbers and related currents below correspond to that displayed in Figure 4: 1: Radio pre-processing period (I MCU_LL and I Standby ), 2: Active radio receive time (I Rx ), 3: Radio Inter frame Space (I TFS ), 4: Active transmit time (I Tx ), 5: Link layer post processing period (I MCU_LL ), 6: Data post processing period, enabled only if data has been received (I MCU_HOST ). I Rx I Tx I TFS I MCU_HOST I VDD I MCU_LL I Standby I idle I on I deep_sleep I on I Standby I sleep 1 2 3 4 5 6 (a) 1_M 2_M 3_M (b) Tconversion Tperiod (c) Figure 4: Current consumption over time for the module ISP091201 (a), for the microprocessor LPC1114FHI33/302 (b) and for the sensing device (c) The values of static current consumption of BLE module ISP091201 for its different states and operations in each connection event are defined in Table 2. t SAME 2013 Conference October 2 & 3, 2013 2
Table 2: Current consumption static values of BLE module Symbol Parameter (condition) Nom Unit I Rx Peak current, receiver active 14.6 ma I Tx Peak current, transmitter active 12.7 ma I TFS Peak current when switching 7 ma between receive and transmit I MCU_HOST Peak current for host processing 5 ma I MCU_LL Peak current for LL processing 3.5 ma I Standby Standby current 1.6 ma I Idle Current drain between connection/advertising events ACI=active mode, 32kHz Osc active 2 ua For the microprocessor LPC1114FHI33/302, each connection event consists of three states and periods, as presented in Figur4-b: 1_M : Read period (I2C communication with sensing device to read measurements), 2_M : Waiting time (related to post-processing period), 3_M : Send period (SPI communication with the BLE to send the measurements). The values of current consumption of microprocessor LPC1114FHI33/302 for its different states and operations in each connection event are defined in Table 3. Table 3: Current consumption values of microprocessor LPC1114FHI33/302 Symbol Parameter (condition) Nom Unit I on Current of active mode 3 to 4 ma I sleep Current of Sleep mode 2 ma All the clocks are active I deep-sleep Current of Deep-Sleep mode All the clocks are turned off 6 ua Finally, for the sensing device, each connection event consists of two states and periods, as presented in Figure 4-c: I Standby : Standby period current, I on : Active period current (I2C communication with microprocessor + conversion time). The values of current consumption of the three sensing devices in each connection event are defined in Table 4. Table 4: Current consumption values of temperature, light and orientation/motion sensors Symbol Nom Unit Temp TMP112 I on 40 ua I Standby 2.2 ua Light APDS-9300 I on 0.24 ma I Standby 3.2 ua Orientation/Motion MMA7660FC I on 0.047 to 0.294 ma I Standby 2 ua The power management consists of minimizing the current consumption of the three subsets BLE, microprocessor and sensor in the two phases of operation. For the first phase, where the BLE, microprocessor and sensing device are active, we must choose the operation parameters of the three subsets that reduce the current consumption to the minimum. We must therefore manage the time and the level of current consumption while maintaining the correct operation according to the intended application. For the second phase, where the subsets are in Standby (Sleep) mode, we must put the BLE and microprocessor in Sleep mode and also deactivate the sensing device. The goal is to reduce the total current consumption near to zero during this long phase. It is important to know that the total current consumption depends also on the data size to be read by the sensing device and to be sent by the microprocessor (after I2C communication) to BLE module using SPI interface, and then from the BLE module to the Master using a BLE wireless connection. In fact, the total current consumption depends on the type and the number of used services defined in the Bluetooth Low Energy protocol [2], for a given application. The transmitted data between the microprocessor and the BLE module are managed by these services. Thus, the number of defined services and the data size of each service will define Data Layer postprocessing period, together with the T X time and, consequently, the level of current consumption. For example, for the temperature node, we have defined just one service with data size of 2 bytes which are the temperature measurements. While for the orientation/motion and light nodes, we have also defined one service but with data size of 3 and 4 bytes respectively, which lead to more current consumption. Clearly the larger the amount of data to be sent, the greater is the Data Layer Postprocessing period. This leads to longer I on times for both the microprocessor and the sending device, which in turn increases overall current consumption. Current measurement methods In this paragraph, we present two complementary methods for current consumption measurements. The first one uses an oscilloscope to measure the current consumption of the BLE module and microprocessor during the connection phase (Phase 1), where the consumption is maximum. As the sensing device consumption is very weak (a few µa), we will see that this measurement method with oscilloscope is not suitable for measuring very low currents during standby phase. Thus, we have developed a second method, based on the principle of charging and discharging a capacitor to measure the total current consumption, including lowconsumption sensing device. In addition, this method allows the measurement of the total current consumption during the Standby-sleep phase (), where current consumption of each subset is very low. In the following, these two methods are described and illustrated by showing the results for the temperature sensor nodes. The measurement results SAME 2013 Conference October 2 & 3, 2013 3
of the three Bluetooth Smart sensor nodes will be summarized at the end of this article (cf Table 7). Oscilloscope method Oscilloscope measurement method is a basic method, which consists of connecting a low-value resistor (0.5 Ohm) to the power feed of each subset BLE, microprocessor and sensing device, as presented in Figure 5. These resistors have been integrated into the wireless miniature nodes. Voltages at each terminal of the resistances are sent to a measurement board thanks to a FPC cable. Figure 5: Electrical Schematic of current measurement for oscilloscope measurement method On the measurement board, an INA195 amplifier (Texas Instruments) with gain of 100 is used to amplify the differential voltage across the terminals of each resistor. So, by measuring the voltage at the output of each INA195, we measure the current consumption ( 0.5 ) of each subset, and then the total current consumption. The high gain of the INA195 allows us to measure relatively low currents. Figure 6 illustrates the two functioning phases of BLE module and microprocessor measured by oscilloscope method for temperature sensor node. To measure the current consumption of BLE module and microprocessor during, it consists in measuring the voltage and duration of each segment of consumption displayed on the oscilloscope, as presented in Figure 7. Figure 6: Two functioning phases, as measured by oscilloscope method µp VCC_ISP091201 VCC_µProcessor VCC_ µp BLE 0.7ms 1.7ms R=0.5Ω R=0.5Ω R=0.5Ω 1.2ms BLE VCC_App Figure 7: Temperature sensor node: current consumption measurements of BLE module, microprocessor and sensing device, as measured by oscilloscope method For the microprocessor current consumption measurements, we have three segments (0.7 ms, 1.7 ms and 1.2 ms), as presented in Figure 7. The corresponding current and charge consumption values in these three segments are presented in Table 5. These three segments correspond to events presented in Table 3. The current and charge consumption calculations associated with each segment are based on the following two equations:. Table 5: Temperature sensor node: current and charge consumption measurements of microprocessor, as measured by oscilloscope method Temperature node: μprocessor LPC1114 Consumption Symbol Time (ms) Voltage (mv) Current (ma) I on (Read) 0.7 170 3.4 2.38 I sleep (Wait) 1.7 100 2 3.4 I on (Data Send) 1.2 170 3.4 4.08 Total 3.6 9.86 Charge (µc) Similarly, to measure the BLE module current consumption, we have 6 consumption segments which correspond to events presented in Table 2. Table 6 presents the corresponding current and charge consumption values of BLE module, as measured by oscilloscope method using the two previous equations. Table 6: Temperature sensor node: current and charge consumption measurements of BLE module, as measured by oscilloscope method Temperature node: BLE module Consumption Symbol Time (ms) Voltage (mv) Current (ma) Charge (µc) I MCU_LL 1.14 230 4.6 5.24 I Standby 1.2 136 2.72 3.26 I Rx 0.24 660 13.2 3.17 I TFS 0.1 480 9.6 0.96 I Tx 0.25 560 11.2 2.80 I MCU_Host 3.64 260 5.2 18.93 Total 6.57 34.36 Concerning the current consumption of the temperature sensing device, it is very low (a few µa). Because of an intrinsic offset of INA195 component used in the oscilloscope measurement method, we can't characterize this low current with this method. This is confirmed by Figure 7, where we can't measure the consumption of the temperature sensing device. Capacitor method To overcome low current limitations of the oscilloscope method and to obtain more accurate measurements, we have developed another measurement method: the capacitor method. This method is based on the principle of charging a calibrated capacitor, then turning off the power supply and finally feeding the sensor node via this charged capacitor. Total current consumption of the sensor node will discharge this capacitor and decrease its voltage. By measuring the voltage drop SAME 2013 Conference October 2 & 3, 2013 4
across the capacitor terminals as a function of time, we can determine the total current consumption and/or the total charge consumption, as presented in the two following equations: This method allows the measurement of low currents (a few µa), particularly to characterize the current consumption of sensing device during the connection phase (), but also to characterize the total current consumption during the Standbysleep (), where the current consumption of BLE module, microprocessor and sensing device is very low. The capacitor used in this method is a ceramic capacitor of about 600 µf. To determine the precise value of this capacitance, we have used an RC circuit to calibrate the capacitance value. In this way, we have connected the capacitor to a resistor of 1 KΩ (tolerance 1%). The time constant of this circuit is given by. Thus, by measuring the time constant on the oscilloscope, we can calibrate the capacitance that will be used to accurately measure the current consumption. The precise value of the capacitance is 530 µf. Figure 8 presents the electrical schematic for total current consumption measurement by the capacitor method. Switch the sensor node for the Standby-sleep phase (Phase 2). Thus, the total charge consumption of the temperature node sensor for the period (T 1s), as presented in Figure 9, is: 530 0.104 55.12 Figure 10 presents the measurements of total consumption of temperature sensor node during the communication phase (). We observe a voltage drop of 88 mv during this phase, so the total charge consumption in the communication phase is: 530 0.088 46.64 These results are in accordance with measurements made for the BLE module and the microprocessor by the oscilloscope method presented in the previous paragraph (Table 5 and Table 6), where: 34.36 and 9.86. The difference between the two measurements is the consumption of the sensing device during the communication phase (), so: 46.64 34.36 9.86 2.42. Figure 11 presents the measurements of total consumption of temperature sensor node during the Standby-sleep (). We observe a voltage drop of 16 mv during this phase, so the total charge consumption in the Standby-sleep phase, is: 530 0.016 8.48 3 V + Module C under - 530 µf test Figure 8: Electrical schematic for current consumption measurements, using capacitor measurement method Figure 9 shows the measurements of total current consumption of the temperature sensor node for the two functioning phases (+), carried out by the capacitor method for a measurement time (Connection Interval) of 1 second. Figure 10: Temperature sensor node: total current consumption measurements during, as measured by capacitor method + Figure 9: Temperature sensor node: total current consumption measurements, as measured by capacitor method In these measurements, the high voltage drop is related to the total current consumption of the sensor node during the communication phase () where there is a high consumption for a short time (6-10 ms). However the low slope of consumption is related to the total consumption of Figure 11: Temperature sensor node: total current consumption measurements during, as measured by capacitor method SAME 2013 Conference October 2 & 3, 2013 5
Thus, the total average current consumption of the temperature sensor node for the period ( 1 ) is: 55.12 1 Life time model/measurements In addition, we have developed a model for calculating current consumption of temperature, light and orientation/motion sensor nodes. This model is based on the data sheets of the components, that is to say, the current consumption values of each subset BLE module, microprocessor and sensing device, as they are given in their data sheets. This model allows the calculation of the average consumption of each subset, the total consumption and the autonomy of sensor nodes for different Connection Intervals and for different types of battery. Table 7 presents model/measurement comparison of the total current consumption and the autonomy of temperature, light and orientation/motion sensor nodes, respectively, for different Connection intervals. Table 7: Model/Measurements comparison for Temperature, Light and Orientation/Motion sensor nodes node Connec Interval Cons Model Cons Meas Auton Model Auton Meas (year) (ms) (µc) (µc) (year) Temp 1000 52.30 55.12 0.31 0.29 2000 62.30 63.6 0.51 0.50 3000 72.30 72.08 0.66 0.67 4000 82.30 80.56 0.78 0.79 Light 1000 65.72 67.84 0.24 0.24 2000 76.92 80.56 0.42 0.40 3000 88.12 93.28 0.54 0.51 4000 99.32 106 0.64 0.60 Orientation Motion 1000 62.72 63.6 0.25 0.25 2000 72.72 72.08 0.44 0.44 3000 82.72 80.56 0.58 0.60 4000 92.72 89.04 0.69 0.72 Regarding this table, we note that the calculation model of current consumption is quite reliable and provides good estimations of consumption and autonomy close enough to the real operation case in the range of measurement time (Connection Interval) 1000 to 4000 ms. Demonstrator Finally, a demonstrator system will be presented during the session of the conference. A short visual presentation of potential applications of these wireless sensor nodes will be made using software developed in house. complementary methods for current consumption measurements have been presented, in addition to a life time calculation model/measurements. The resulting sensor node has overall PCB dimensions of 18 x 29 x 6 mm 3 with a life time between 0.29 year and 0.79 year using 3V coin cell battery CR1632. Acknowledgements This work was supported by the French National Research Agency (ANR) project GRECO bearing reference ANR-2010-SEGI-004-04. References [1] P. CIAIS, C. El HASSANI and J. MOLON : Ultra Low Power System-In-Package Module For Bluetooth Low Energy Applications, SAME 2010 Forum, Sophia-Antipolis, France, 6-7 Oct 2010 [2] Bluetooth Core Specification Version 4.0, 17 th December 2010, http://www.bluetooth.com/specification Documents/Core_V40.zip About the Authors Atef AL NUKARI PhD graduate in Electronic speciality RF/Modules from the University of Nantes in France. 14 years experience in RF and Microwave design with specialization in RF compact power transmitters optimized in efficiency and consumption. antennas. Pascal CIAIS PhD graduate in Electronic speciality RF/Antenna from the University of Nice Sophia- Antipolis in France. 12 years experience in RF and Microwave design with specialization in the design, optimization, analysis and measurement of miniature Conclusions This paper has described the design and the performance of low power Bluetooth Smart sensor nodes. In particular, the optimization of the current consumption is clearly demonstrated. Two SAME 2013 Conference October 2 & 3, 2013 6